Tory EvansThe Wrong Person Is Being Sold AI Right Now
March 24, 2026
There is a pitch deck somewhere - probably several hundred of them - that opens with a slide titled "AI-Powered [Something]" and ends with a request for a meeting with the CFO. The vendors making these decks believe that if they can get in front of the person controlling the budget, they've won. They are wrong about this. They are wrong in a very specific and expensive way.
The CFO doesn't care about AI features. Not because they're behind the times. Not because they don't believe in the technology. It's because their job is to measure outcomes, not evaluate features - and right now, the AI market is still mostly selling features. Meanwhile, the team that actually touches purchase orders, contracts, supplier relationships, and spend data every day - procurement - is the one that should be having these conversations, and in most companies they're either being left out entirely or handed watered-down tooling after the CFO signs off on something they half-understood.
This divide is real, it's documented, and it's costing businesses money in both directions. Let me explain what's actually happening here.
The Adoption Gap Is Not About Budget. It's About Who's Driving.
I want to start with a number that stopped me cold when I found it. While 85% of procurement leaders express optimism about AI's potential efficiency gains, 61% of procurement teams have yet to implement AI into their workflows. That's not a skepticism problem. That's an execution problem. And I'd argue it's an ownership problem.
When AI buying decisions get elevated to the CFO level before the procurement team has had a real chance to evaluate the tools, you end up with exactly this: enthusiasm at the top, paralysis in the middle, and nothing deployed at the bottom. Where CFOs have adopted generative AI, they have often reached for tools from their existing vendors, rather than adopting net new, AI-native solutions. Which makes complete sense from a CFO's perspective - minimize risk, stay within the existing spend envelope, check the box. But procurement teams working on actual source-to-pay workflows have different, more specific needs that a bolt-on feature from SAP or Oracle doesn't always meet.
The result is a function that's being asked to do more with less but isn't being given the right tools to do it. McKinsey's research shows that the procurement function's spending managed per full-time equivalent is 50 percent more today than five years ago. The workload went up. The headcount didn't. And most companies responded not by fundamentally rethinking how procurement operates, but by asking people to work harder - occasionally with a new dashboard.
What the CFO Actually Sees
I'm not saying CFOs are wrong to be cautious. I'm saying they're cautious about the wrong things.
79% of CFOs surveyed indicate their AI budget will increase in 2025, and 94% indicate that generative AI can strongly benefit at least one activity area within the finance organization in the next 12 months. But here's the catch: 71% of CFOs surveyed are not currently using generative AI in their finance and accounting function. They believe in it. They're budgeting for it. They're just not actually doing it yet.
This is the CFO operating in their natural habitat. They evaluate. They monitor. They wait for the ROI case to solidify before they commit. That's appropriate behavior for someone who has to answer to a board. The problem is that when this same disposition gets applied to procurement's AI tooling - when the CFO becomes the de facto decision-maker for what the sourcing team gets to use - it creates a bottleneck that procurement can't afford.
Chris mentioned to me last week that he'd been trying to get sign-off on a new spend analytics tool for about four months. The demo had gone well. The use case was clear. But it kept getting pulled into a broader "AI strategy review" at the executive level and nobody could tell him when that review would actually conclude. He wasn't frustrated. Chris is never frustrated. He genuinely couldn't understand why anyone would have a problem with the situation. I envied him deeply.
What Procurement Actually Needs (And Is Missing)
The procurement use case for AI is not abstract. It's not "enhanced insights" or "better decision support." It is extremely specific and the returns are measurable in ways a CFO should theoretically love - but only if someone with procurement context is doing the evaluation.
According to a poll of CPOs, the top use cases for generative AI in procurement are spend analytics and dashboarding (53%), RFP/RFQ generation (42%), and contract summarization and key terms extraction (41%). These aren't moonshot applications. These are things procurement teams do manually right now, every week, that chew through hours and create real risk from human error. AI-powered platforms have transformed contract lifecycle management - beyond simple document storage, these intelligent platforms automatically extract key terms, flag renewal dates, and identify opportunities for consolidation or renegotiation across the contract portfolio.
The financial impact when this actually gets implemented is not small. In one pharmaceutical example, an AI audit recovered $10 million in missed value in under a month. That's not an efficiency gain. That's found money. And it happened because a procurement team had tools capable of seeing what the humans couldn't. That kind of outcome doesn't come from a CFO picking a vendor at arm's length. It comes from procurement teams that understood their own workflows well enough to evaluate what an AI tool could actually do inside them.
The market is growing to meet this demand. The global AI in procurement market was calculated at $3.32 billion in 2025 and is predicted to increase to approximately $39.20 billion by 2035, expanding at a CAGR of 28%. That trajectory only makes sense if procurement teams become active buyers and evaluators - not passive recipients of whatever the finance suite vendor bundles in.
The Scale Problem Nobody Talks About
Here is the uncomfortable truth buried in the optimism: almost nobody has actually scaled this yet. 94% of procurement executives use generative AI at least weekly - but only 4% achieved large-scale deployment in 2024. That gap between personal usage and organizational deployment is where the real story is.
Individual procurement people are using AI. They're using it in their browsers, in their personal ChatGPT accounts, in whatever tools their teams have quietly adopted. But global CPOs observe that the systematic use of generative AI has not yet been rolled out across their teams. The enterprise infrastructure isn't there. The data governance isn't there. And in a lot of cases, the internal champion who could push it through doesn't have the organizational authority to do it without CFO approval.
This is the cycle that kills transformation in mid-market companies especially. The people who understand the problem can't authorize the solution. The people who can authorize the solution don't fully understand the problem. And so nothing moves until someone decides to force the issue or a competitor does it first.
I thought about this a lot around 11pm last Tuesday, parked in the garage off Fifth because the wifi from the office building still reaches the lower level. I was working through a client's software audit - they'd asked me to look at where they were bleeding budget - and I kept finding the same pattern. They had a CFO who had approved a big ERP with AI modules built in. The procurement team had never been fully trained on the modules. The modules were renewing automatically every year. Nobody was using them. The CFO thought they were. Growth sometimes looks like finding out you were paying for something that wasn't working, which is a form of learning I've become very familiar with.
If you want to get serious about this kind of audit for your own stack, we did something similar across a full software stack a while back and the findings were genuinely surprising - we documented what happened here.
The Vendor Side of This Is Changing Fast
To be fair to the AI vendors, the better ones have figured out that selling to CFOs doesn't work if procurement never adopts the tool. Deloitte's 2025 CPO survey found that "Digital Leader" procurement teams achieved 3.2 times higher ROI on AI investments than peers. That number is starting to appear in sales pitches, which is good - it's the right framing.
The more interesting development is what's happening at the platform level. In September 2024, IBM announced the launch of watsonx for Procurement, a new suite of generative AI agents designed to automate complex sourcing events, from RFP generation to advanced supplier negotiation simulations. In November 2024, SAP and NVIDIA revealed a strategic partnership to integrate NVIDIA's AI Enterprise software into the SAP Ariba platform, aiming to accelerate predictive analytics for supply chain risk. These aren't feature announcements. These are architectural bets that procurement, not finance, is going to be the primary beneficiary of enterprise AI.
The companies that are moving fastest are the ones treating procurement as the wedge, not the afterthought. According to ISG's 2025 State of Enterprise AI Adoption research, supplier risk assessment and monitoring has emerged as one of the most successfully deployed AI use cases in procurement, with 58% of implementations already in production. That's a remarkable production rate compared to most enterprise AI applications. It's high because the use case is clear, the data exists, and the ROI is measurable. Procurement teams, when given actual authority over their own tooling decisions, make good calls.
Derek disagreed with me on this in the group chat. He thinks the CFO staying involved is a "checks and balances thing" and that procurement teams will just buy tools that make their lives easier without thinking about total cost of ownership. He's not entirely wrong. But he also thinks the prequel trilogy is better than the original Star Wars and I've made my peace with the fact that Derek and I will not always agree on matters of strategic importance.
The McKinsey Number That Should End the Debate
I'm going to give you one more data point and then tell you what I actually think is going to happen.
McKinsey's long-running Global Procurement Excellence benchmark shows that companies with advanced procurement operating models enjoy five percentage points higher EBITDA margins than peers. Five points. That is not a rounding error. That is a structural competitive advantage that compounds over time, and it correlates directly with how much operational autonomy and tooling quality the procurement function has.
The CFO looking at the ROI case for AI should want to fund procurement's AI adoption more aggressively than almost any other function. But they have to stop treating procurement's tool selection as a feature procurement review and start treating it as a margin improvement program. Those are different conversations with different ownership and different timelines.
Linda sent me an article about this topic last week with a note that said "Gerald thinks procurement is undervalued in most organizations and has said so for years." Gerald has been married to Linda for 31 years and apparently has strong opinions about procurement strategy. Gerald is right.
What I Think Happens Next
I think the CFO role in AI procurement decisions shrinks over the next 18 months. Not because CFOs lose influence, but because the ROI case for procurement AI has become undeniable enough that it stops requiring executive-level justification for every tool purchase. Over half of CPOs - 53% - have increased their source-to-pay technology budgets compared to the previous year, a notable jump from 30% in 2024, suggesting a strengthening confidence in digital solutions. Budget authority is following conviction downward through the org chart.
McKinsey's analysis suggests this shift could result in the procurement function being 25 to 40 percent more efficient, while repurposing team activity from routine tasks to strategic decision making. When efficiency gains hit that level, they show up in EBITDA before the annual review cycle. The CFO doesn't need to evaluate the AI features at that point. They see the number move and they get out of the way.
The businesses that win this transition are the ones that let procurement lead the evaluation process now - before the efficiency gap between them and their competitors becomes visible in the market data. 79% of organizations report difficulty in purchasing B2B software and services, leading to mismatched investments and inefficiencies. A lot of that difficulty is structural. It's what happens when the people closest to the problem aren't the ones choosing the solution.
If you're in procurement right now and you're waiting for the CFO to hand you a tool, stop waiting. Build the business case around margin impact and risk reduction - the two things that actually get a CFO's attention - and then own the evaluation yourself. The tools are there. The data on what works is there. The only thing missing is the internal positioning to make it happen.
I'm aware this is easy to say. I'm aware that organizational change is slow and that telling someone to "own the evaluation" doesn't automatically give them the authority to do it. I know something about wanting to move forward and finding the mechanism temporarily unavailable. But the companies that figure this out in the next 12 months are going to be very hard to compete with. That's not motivational-text optimism. That's just what the data says.
And if you're building out your sales intelligence to make a stronger case to internal stakeholders, a good place to start is understanding what the best tools actually look like - we've done the work on the sales intelligence landscape here. And if the bottleneck in your organization is genuinely at the CRM level - the kind of thing that makes procurement and finance feel like they're working from different planets - the CRM comparison we built is worth a look as a diagnostic before you try to add AI on top of a disconnected stack.
The AI features don't matter if the right person isn't evaluating them. That person is in procurement. It's time to act like it.